The boiler feedwater line to the reactor coils failed after 24 years. It had been patched several times but no records were kept. The superintendent noted that valve flange bolts mysteriously fell apart. Yet, one manager dismissed concerns because he thought 24 years was a pretty good service life.
Cutting out a section before the line failed would have been nice but production had a “run to failure” attitude. Now, an elbow looks like a spaghetti strainer with 85-psig water gushing out; a clamp has bought some time but what next? Put in a more expensive alloy? Make the pipe wall thicker? These are common impulses.
Instead, I heeded the words of Gene Kranz, the chief flight director during the Apollo 13 mission: “Work the problem. Don’t make matters worse by guessing.”
I laid out a plan: evaluate how boiler chemicals are added and monitored; collect history on the pipe; determine the pipeline fluid chemistry upstream and downstream; do a complete metallurgical study of the failed pipe and fresh pipe around it; and perform a start-to-finish inspection of all 550 ft of the pipeline. One critic complained that the boiler feedwater pump, tank and boiler line didn’t show any corrosion; I wasn’t deterred: one line at a time.
I collected data on the boiler feedwater chemicals. As often occurs, the additions were based on shift lab analysis — definitely not steady state.
The chemical supplier agreed to test water samples and predicted that chlorine in the city water was the culprit. Sure enough, the lab report indicated chlorine is a problem. However, as I then pointed out, the side reaction of ClO- (hypochlorite) with sodium sulfite (the O2 scavenger) can produce HCl and, more importantly, deprive the system of scavenger. The hypochlorite reaction is more exothermic than the scavenger reaction. So, I insisted on waiting to see what the metallurgy study revealed.
[pullquote]
The metallurgy study involves destructive testing. Samples are dye-tested with acid for inspection of the metal crystals and spaces between them. X-raying of other samples determines precisely what’s left in the metal and what corrosion has washed away. In addition, tensile testing shows how the mechanical strength of the metal compares to literature values.
The inspector ran positive material identification, i.e., x-ray fluorescence, and ultrasound thickness (UT) measurements on spots from beginning to end of the pipeline. I marked up a Google map of the pipe run because it was outside; as a general rule, be skeptical of layout drawings. The map showed the sample points for the inspection. After insulation was removed, each pipe sample was marked with a number. The results came back showing a variety of pipe schedules, materials and thicknesses.
Compensating for temperature is important in UT measurements. ASTM E797-95 notes that sound velocity changes by 1% for carbon steel and austenitic steels per 100°F difference between the calibration temperature, say, 68°F, and the wall temperature, typically measured with an infrared gun. So, for a pipe wall thickness measurement of 0.18 in. at 250°F, the actual thickness is 0.176 in.
The hoop stress calculation showed failure wall thickness was just about the wall thickness measured at the hole. I decided to do a quick calculation of the remaining pipe life and corrosion rate: CR = (toriginal – tmeasured)/spent life of pipe × 1,000 mills/inch = mills/year; then, (tmeasured – tfailure) × 1,000/CR = remaining absolute life (with no safety factor).
Another reason why you have an inspector onsite is to determine how much of this cancer must be cut away. You don’t want to weld onto thin pipe. Of course, because thickness is a critical measurement, you might want to have a UT gun onsite not only to confirm tie-ins but also to determine pipe schedule and tank wall thicknesses.
This crude first step made two things clear: 1) the remaining life was at least five years, not nearly as short as feared by those wondering how they’d find $130,000 during a pandemic; and 2) the corrosion might be showing a pattern with shorter life upstream than downstream.
A common problem in maintenance is jumping the gun and doing repairs before they really are required.
